Development Of Micro Near-infrared Spectroscopy Based On Android System And Its Application In Food Quality Detection

The near infrared spectral analysis technology has become one of the simple process, rapid, non-destructive, chemical information-rich, pollution-free and simultaneous multicomponent technologies and has been widely used in agricultrue, food, tobacco, petrochemical and so on. With the development of science and technology, and the needs of the practical application, the near infrared spectrometer presents the trend of miniaturization and portability, take JDSU of America for example, the company launched MicroNIR-1700 near infrared spectrometer in 2012. Compared with China, however, due to the complexity of the near infrared spectral analysis system, there is few near infrared spectrometer in the country, and most of which are made from foreign companies and the price of which limit the scope the of it’s application.Therefore, the internal market needs a cost-effective miniature near infrared spectrometer, which meets the general requirement of general application miniature near infrared spectrometer.This article is based on the study of Hamamatsu C11708 MA mini near infrared spectrometer, the instrument were made by working group including design, integrated and development. Frist of all, as the core hardware of the near-infrared spectroscopy system, we evaluated the basic properties of the near-infrared spectrometer and carried out feasibility test. Secondly, determined the overall system software application framework based on modular and practical design principle. Thirdly, designed and developed the system application software in the Windows system and mobile terminal based on the.Net and Android platforms respectively. Finally, conducted some practical verification experiment to test the practicability of the near-infrared spectrometer. The main contents are as follows:1. Introduced the technical parameter of C11708 MA near-infrared and analysed the working principle of infrared detectors from hardware system; depending on the differences of sample, designed testing accessories applied to liquids, solids and powdered samples. We accessed the performance of C11708 MA near infrared spectrometer from 3 indicators----the output signals, absorbance repeatability and baseline, and analysed the impact of the integration time on the frist three aspects. Based on the results of the performance evaluation, we could draw a conclusion that applied to the method of optimization of control parameters, namely for the 16 bit ADC of the detector of near infrared spectrometer, the user should consider two factors: the output signal level is in 20000~40000Counts and single sample measurement time is within 1~2s. In this case, increased the integration time can achieve better results.2. According to near-infrared spectroscopy detection process and model building process, we designed the miniature NIR spectrometer application framework. In order to meet the needs of a large number of spectral data acquisition and model building for related researchers in near-infrared spectral research, we developed a software using C# based on Microsoft’s.Net platform. Meanwhile, used Java languages to develop spectrometer application app on the Android platform, realized the functions of the basic control of near-infrared spectroscopy, spectral acquisition, model load and save data, facilitated users to quickly detect and analyse the chemical index of samples, reflected the faster and more efficient advantage of near infrared spectral analysis technology.3. Used Android smart phones as a control terminal of system, we developed a simple and good human-computer interface with Java open source feature and a powerful API function provided by Google Company. Realized the functions of spectrometer hardware controlling, data saved, model download, model file called(the model file was set up by NIRSA data processing system, the working group developed) and real-time detection of sample spectra and component.4. Introduced the technology process of near infrared spectroscopy analysis. In order to study the practicability of the near infrared spectrometer, we used the ethanol of different concentration and apple from Jinan orchard as test objects, ethanol and soluble solids content as the specific detection quota. In the ethanol experiment, 35 ethanol samples were used to build calibration models by stepwise linear regression(SWR), principal component regression(PCR) and practical least squares regression(PLS) method respectively, the rest 15 samples were used to forecast model. The results showed that the model made by PLS regression method was better, and it’scorrelation coefficient of prediction was 0.9988, the average prediction error was 1.959%, the maximum error was 4.909%. Studyed on solid90 apple samples, 63 samples were used to build calibration model by partial least squares regression(PLS), the rest 27 samples were used to predict the model and removed the 2 samples by using Manhattan distance discrimination. The final built model showed that the prediction coefficient was 0.9283, the average error was 2.701%, and the maximum error was 9.179%. The established models were stable and reliable, and can meet the practical application.